Down is easy. Degraded is where monitoring earns its keep.

Any monitoring tool can tell you a site is down: the request fails, the alert fires, everyone knows. The incidents that actually cost money look different: the checkout that takes nine seconds, the API that answers 200 OK after a four-second think. Nothing is 'down'. Customers leave anyway.

A third state between up and down

That's why every Uptimix monitor has a degraded threshold: a response-time budget in milliseconds. A response that arrives but blows the budget marks the monitor degraded, visibly distinct from up, and alertable on its own if you want it to be.

The threshold is per monitor, because a static landing page and a search endpoint don't share a definition of slow. Start with something forgiving, watch the response-time chart for a week, then tighten it to just above your real p95.

Slow creep is a prediction

The most valuable thing about tracking degradation is what it tells you about tomorrow. Memory leaks, filling disks, and swelling tables rarely jump straight to an outage; they drift there, one slightly slower response at a time. A degraded streak is often the only warning you get, which is exactly why our upcoming predictive insights start from response-time drift.